SOLVENT
EXTRACTION N°1
English Translation by
N. L. Ricker, University of Washington
May, 2005
ATELIER INTER UNIVERSITAIRE DE GENIE DES PROCEDES
5 Rue Paulin Talabot 31106 TOULOUSE Cedex 1 -
Tel : 05-34-61-52-33
-
Fax : 05-34-61-52-53
- AIGEP@ensiacet.fr
I. Background1
Solvent extraction (or liquid-liquid extraction) is an isothermal unit operation involving
two immiscible or partially miscible liquid phases in direct contact. The goal is to transfer
one or more solutes from one liquid phase to another, thereby achieving a separation.
A single liquid extraction stage involves three steps, illustrated in Figure 1 for the case
of one transferring solute :
•
Mixing of the liquids, one of which (the feed mixture) contains the solute. The
other (the solvent) has a higher affinity for the solute.
•
Solute transfer from the feed mixture to the solvent. The direction of transfer is
such that the phases approach an equilibrium condition. In an ideal equilibrium
stage, the phases achieve equilibrium.
•
Phase separation or disengagement, usually driven by a density difference.
The solvent phase leaves as the extract, which is now richer in solute. Ideally, the extract
would contain all of the feed solute, but phase equilibrium conditions are almost never that
favorable. Thus, some solute remains in the raffinate product, but the raffinate is leaner in
solute than the feed.
feed mixture
extract
equilibration
solvent
raffinate
mixing
phase separation
Figure 1 : Schematic of a solvent extraction stage.
In order to achieve a high degree of solute removal from the feed mixture, multiple
stages are used (as in absorption and other separations driven by phase equilibrium effects).
Such stages can be individual physical units, or they can be combined in a column
1
Perry’s Handbook, 7th Ed., Ch. 15 covers liquid-liquid extraction in more detail than is
possible here.
-2-
configuration. In either case, the phases flow from stage to stage counter-currently as in
absorption.
The solute transfer rate depends strongly on the interfacial area available for mass
transfer. To encourage a high transfer rate, liquid extraction devices are designed to mix the
phases intimately, leading to a high interfacial area.
II . Ternary liquid-liquid phase equilibria
1 – Partition Coefficient
Assume that a solute B has equilibrated between two partially miscible liquid phases,
one of which is mainly chemical A, and the other mainly chemical C. Let the B concentration
in the equilibrated phases be x and y. The partition coefficient is the ratio of these
concentrations, for example : m = y / x . The definition is somewhat arbitrary : various
concentration units are used (e.g., mass fraction, mol fraction) and the ratio could be inverted,
so be careful and consistent.
2 – Ternary LLE Diagram
It is convenient to display ternary liquid-liquid equilibria (LLE, at constant temperature
and pressure) on an equilateral-triangle ternary diagram, an example of which appears below.
Each triangle vertex represents a pure chemical (species A, B, and C in this example). The
composition of a mixture, such as the point M in Fig. 2, is obtained by drawing projections
parallel to each base through the mixture point, as shown. For example, to read the % B for
point M, calculate the ratio CM B / CB , where CM B represents the distance between points C
and MB, etc. Thus, point M is about 40% B, 40% A, and 20% C.
-3-
100% B
B
MA
N
M
MB
A
100% A
C
MC
100% C
Figure 2 : Ternary Diagram
Point N represents a mixture of roughly 50% A and B, with no C. Notice that a line
drawn through N and the pure C vertex also passes through M, i.e., the three points are
colinear.
This implies that we could obtain mixture M by mixing the appropriate amounts of
mixture N and pure C. The inverse lever-arm rule gives the required proportions :
amount of N length MS
=
amount of S length NS
3 – Phase Boundaries
B
1
F
f
3
4
R
2
M
E
A
f"
f'
S
Figure 3 : Ternary LLE diagram showing phase boundaries and equilibrium tie lines
-4-
Figure 3 shows a ternary system in which two of the chemicals (A and S) are partially
miscible, and the other two pairs (A and B, and B and S) are totally miscible. Zone (1) is a
single-phase region. Points inside the curve f ′ − f − f ′′ , i.e., zone (2), are mixtures that
separate spontaneously into two equilibrium liquid phases. One phase will have a
composition somewhere along the phase boundary f ′ − f , and the other will be somewhere
along the other phase boundary, f − f ′′ . The line joining such a pair of phases in equilibrium
is a tie line.
For example, point M is within zone (2), so it will separate into two equilibrium phases.
Suppose the line going through M is a tie line, and that it crosses the phase boundaries at
points R and E, i.e., R, E, and M are colinear and on a tie line. Thus, R and E are the
equilibrium phases obtained when M separates spontaneously. Also, the inverse lever rule
gives the phase proportions, as follows :
amount E/amount R = RM ME
Points along the line f ′f ′′ represent mixtures of A and S (and zero B) that will split into
equilibrium phases having compositions f ′ and f ′′ . In other words, f ′ and f ′′ give the
mutual solubilities of A and S. As one adds B, the tie lines change slope and become shorter,
eventually disappearing at the critical point, f .
Most literature data give a few tie-line composition pairs. You must either interpolate
as best you can to obtain other tie lines, or fit the data to a thermodynamic model, then use the
model to interpolate. The former approach is better unless you expect to re-use the model for
another purpose.
4 – Staged separation
B
B
Y
BE
R
A
f'
f
f
f
M
f" S
Y(E)
M
R
E
E
f"
f'
S
X(R)
X
Figure 4 : Three representations of a ternary system with limited miscibility.
-5-
BR
Imagine a point F along the line AB in Fig. 4.
Such a point cannot separate
spontaneously into two liquid phases (A & B are completely miscible). Suppose that points F,
M, and S are colinear. Then mixing the appropriate proportions (inverse lever rule) of pure S
and mixture F will yield mixture M. Finally suppose R, M and E are colinear and on a tie
line. Then M will separate into mixtures R (the raffinate) and E (the extract) the latter being
more rich in B. In general, the location of M governs the compositions of R and E.
Figure 4 shows two other common ways to represent ternary LLE. The right triangle
diagram in the middle shows the same data on an A-free basis. The vertices B and S
represent 100% B and 100% S respectively. The other vertex is 0% B, 0% S. You read any
point on this diagram as a certain %B and %S, with the balance, % A = 100 - %B - %S. In
effect, we have reduced the ternary system to a binary representation. R and E represent a
raffinate and extract in equilibrium, as before. Let x be the % of B in the raffinate, and y the
% of B in the extract. Then the locus of all possible extract/raffinate combinations generates
an x-y diagram like the one on the right-hand side of Fig. 4. This can be used to step off
equilibrium stages, just as in absorption or other equilibrium-stage unit operations.
III . Liquid Extraction Processes
1 – Single Stage
Given a quantity of solution F containing solute B, one adds a quantity of solvent S (not
necessarily pure). Assume that M is the resulting mixture point colinear with F and S, and
that it is inside zone (2). Mixture M separates to form extract E1 and raffinate R1 (see Fig. 5).
R1
F
Etage 1
S
E1
Figure 5 : Single-stage extraction.
2 – Multi-stage, cross-flow
If a single extraction stage doesn’t recover enough of the solute, raffinate R1 can be
contacted with additional solvent, forming another extract E2 and raffinate R2. This can be
-6-
repeated for n stages, as shown in Fig. 6. The n extracts, E1 to En, can be combined to form a
single product.
E1
F
Etage 1
S
E3
E2
R2
R1
Etage 2
En
R3
Etage 3
S
Rn
Rn-1
Etage n
S
S
Figure 6 : Multi-stage extraction, cross-flow of solvent.
IV. Multi-stage, counter-current extraction
The process shown in Fig. 6 can achieve a high recovery of B from mixture F, but at the
cost of a high solvent consumption and a combined extract product that is dilute in B.
Although the solvent, S, is chosen to make a subsequent S/B separation easy (e.g., by
distillation), it is economically advantageous to minimize solvent use. A counter-current
arrangement does this. See Figure 7. If these are equilibrium stages, the streams exiting stage
i, Ei and Ri, are in equilibrium, and must correspond to a tie-line pair on the LLE diagram.
-7-
F
y
1
xF
E1= E
1
R1
E2
2
R
2
R
E3
i-1
Ei
i
Ri
E i+1
R n-1
En
n
Rn= R
x
S
y
n
S
Figure 7: Counter-current, multi-stage liquid extraction process
Mass Balance (steady state)
For a system boundary around the entire column, the steady-state conservation of total
mass and solute gives
Total mass :
mF + mS = mE + mR = mM
Solute B:
xF mF + xS mS = y1`mE + xn mR = xM mM
where mi is the mass flow rate of stream i, and xi or yi is the solute mass fraction in stream
i. These equations have 8 unknowns. They can be solved for 2 unknowns if the other 6 are
specified. In the experiment, the mass rates and compositions of the feed and solvent will be
specified. Setting two other variables, e.g., xn and mR , allows the equations to be solved.
-8-
The mass balance can also be solved graphically. Let a point M on the ternary diagram
represent a hypothetical homogeneous mixture obtained by combining the two process feed
streams. At steady-state, it must also represent the mixture one would obtain by combining
the two product streams. Point M must be colinear with points F and S, and with points E and
R. Thus, once F and S have been located, the solvent-to-feed ratio, mS / mF (and the inverse
lever-arm rule) determines the location of M (see also its determination using the two mass
balance equations above). Then if point R is chosen (it must be somewhere on a phase
boundary), point E is determined by the point where a line through R and M crosses the other
phase boundary. Alternatively, chosing point E determines point R.
Once point E is known (= E1), if we assume ideal equilibrium stages, the equilibrium tie
line through E1 determines the location of R1. Next we would like to determine E2 (the
passing stream for R1). To do so, define a mass balance envelope that goes around one end of
the column (say the end where E1 exits) and cuts through the passing streams of interest (R1
and E2 in this case). In general, for the passing streams between stages i and i+1, we get
Total mass :
mRi − mEi +1 = mF − mE = mD
Solute B:
xR i mRi − yEi +1 mEi +1 = xF mF − yE mE = xD mD
where D is a hypothetical mixture composition representing the difference between the
upward and downward flowing phases. There is no introduction or removal of material at an
intermediate stage in the column, so this same D point must apply for each set of passing
streams. To locate it graphically, note that F, E and D must be colinear, and S, R and D must
be colinear. Thus, point D is the intersection of a line through F and E with a line through R
and S. NOTE : D is hypothetical and usually lies outside the ternary diagram, as in the
example in Fig. 8. If so, at least one of the compositions defining D is negative.
Once D has been established it is possible to step off stages as follows :
1. To determine Ei+1, the unknown passing stream between stages i and i+1, Draw
a line through the known point Ri and D. Ei+1 must be on this line, and it must
also be on the phase boundary opposite Ri. Thus, Ei+1 is determined.
2. Use a tie line from Ei+1 to locate point Ri+1.
3. If stage i+1 is not the last, i.e., if xi +1 > xR , repeat steps 1 and 2.
-9-
In general, unless you have been very fortunate in your choice of E and R, you won’t hit xR
perfectly with an integer number of equilibrium stages. You can determine how much you
would overshoot the assumed xR in one more stage, then use linear interpolation to determine
the hypothetical fractional needed to achieve xR .
B
E=E1
E2
F
R1
E3
R2
D
M
P
Rn=R
S
Figure 8 : Graphical determination of the number of equilibrium stages.
The minimum solvent-to-feed ratio needed to achieve a particular separation
corresponds to a choice of R and E such that one of the lines through point D coincides with a
tie line. This represents a pinch condition for which compositions no longer change from
stage to stage. In other words, an infinite number of stages would be needed to achieve the
separation. The pinch can occur at one end of the column or somewhere in the middle. If the
solvent is a pure material, as in Fig. 8, the pinch can’t occur at the end where the solvent
enters (there will always be a mass transfer driving force at that end).
To determine whether the pinch occurs at the other end, locate (by interpolation) a tie
line that, when extended, would go through point F. The E defined by this tie line is a
potential condition of ( mS / mF )min because the line through E and F also defines point D. It
is the true pinch location if you are able to step stages from the other end of the column until
you reach the pinch.
Another check is to plot the equilibrium and operating lines in the format of the rightmost graph in Fig. 4. The operating line is the locus of all passing stream compositions,
which can be obtained by radiating lines from the D point, as in Fig. 8. If the operating line
touches but does not cross the equilibrium line, you have determined the true pinch condition.
If you have assumed a pinch at the end but this causes the operating line to cross the
- 10 -
equilibrium line, the pinch must be in the middle, and you will have to find it by graphical
trial and error.
In practice, one chooses mS / mF such that
1. mS / mF > ( mS / mF )min , and
2. The extraction factor is between about 1.5 and 2. The extraction factor is
m
ε =m E
mR
where m is the partition coefficient.
If the extraction factor is too small, a large number of stages will be needed to achieve the
desired separation. A large value implies that the process is using a large solvent flow rate,
which reduces capital costs but increases operating costs. The above guidelines are a rule-ofthumb that usually results in an economical operation.
- 11 -
V. Experiment
1 - Objective
The goal is to recover acetone from a cyclohexane/acetone mixture using water as the
solvent. The available equipment is a plate-type extraction column with a Rushton-type
mechanical mixer between plates to disperse one phase within the other (thereby increasing
mass-transfer surface area). Flow is counter-current, by gravity (phase density differences).
2 - Materials
Figure 9 is a schematic. Numbered items there are as follows :
1 : Light-phase feed vessel (light = organic).
2 : Light-phase feed pump, positive displacement. *
3 : Heavy-phase feed vessel (heavy = aqueous).
4 : Heavy-phase feed pump, positive displacement. *
5 : Heavy-phase feed line.
6:
Upper glass decanter for disengaging the exiting light product from the incoming
heavy feed.
7 : Rushton-type extraction column.
8 : Sampling points for light and heavy phases.
9:
Lower decanter for disengaging the exiting heavy product from the incoming light
feed.
10 : Heavy phase product line arranged in a u-tube to keep the column from emptying.
11 : Heavy product storage vessel.
12 : Light product line.
13 : Light product storage vessel.
14 : Turbine agitation control **
15 : Heavy phase flow meter.
16 : Light phase flow meter.
*
Regulate pumping rate by adjusting the frequency and amplitude of positivedisplacement pulsations.
** The Ruston turbine RPM can be adjusted manually.
- 12 -
14
12
6
16
8
7
11
5
13
8
15
10
7
8
3
1
2
9
4
Figure 9 : Schematic of the Liquid-Liquid Extraction System.
- 13 -
3 - Conditions
- Feed solution is cyclohexane with dissolved acetone as the solute.
- Solvent is water.
- Initial feed solution rate settings : pulsation 60, course 60.
- Initial solvent rate settings: pulsation 50, course 50.
- Initial turbine rotation : position 4.
Solvant
(eau)
Raffinat
(cyclohexane +
ε acétone)
(acétone)
Extrait
Alimentation
(eau +
acétone)
(cyclohexane +
acétone)
Figure 10 : Overview of the Experiment.
4 - Procedures
•
•
•
•
•
•
Verify that the two feed vessels contain enough solution to perform a run.
Turn on the power at the control panel. Get help from the instructor the first time.
Set the pumps to the desired rates (piston amplitude and pulsation frequency).
Turn on the turbine and set/note the rotational speed.
Verify that the product vessels are receiving liquid.
Use the valve on the heavy product line to regulate the location of the phase
disengagement interface. If the heavy phase is the dispersed phase, this interface will
be at the bottom of the column, in decanter (9). If the light phase is dispersed, the
interface will be in the other decanter (6). The goal is a steady interface location.
- 14 -
•
•
•
•
Take a sample of the feed solution (from the feed tank, not the pump exit).
About every 15 minutes, take an extract sample and analyze it (gas chromatography,
GC). Wait for steady state.
Use GC to determine the steady-state compositions of both feeds and both products.
Measure feed and product rates. For the feeds, use the graduations on the feed tanks.
For the products, use the rotameters and/or the amount collected in a small container
over a measured time period.
6 – Variables
•
•
•
Change the solvent rate to : course 30, pulsation 30. Hold other settings at their initial
values (given in section 3).
Use turbine position 2, with the feed and solvent rates at their Section 3 conditions.
Change the feed solution rate to : course 40, pulsation 90. Hold other settings at the
values given in section 3.
7 – Report requirements
•
•
•
•
•
•
•
•
Draw the ternary diagram (see data below).
Show points corresponding to the feed solution, raffinate, and extract.
Determine graphically the number of equilibrium stages provided by the column and
the HETP.
[ xsolute / xinert ]extract
Calculate the selectivity, S =
, where cyclohexanone is the inert
[ xsolute / xinert ]raffinate
in the extract phase, and water is the inert in the raffinate phase.
Use the inverse lever rule and your diagram to calculate the solvent/feed ratio.
Compare to the actual.
Check the component mass balances.
Calculate the extraction factor and % solute recovery from the feed.
Determine and comment on the effects of the system variables.
- 15 -
VI – Data
The following data give % by mass for the ternary cyclohexane/acetone/water system at
20 C, 1 bar. Use the first table to plot the phase boundaries. Use the second table to locate
tie-line pairs on the two phase boundaries.
1. Phase boundary points
Cyclohexane (%)
A
67.53
47.81
38.88
29.25
20.24
17.02
14.28
11.76
5.19
1.73
0.98
0.40
0.20
Acétone (%)
B
31.98
50.04
57.52
64.87
69.52
70.78
71.43
71.79
69.25
60.32
55.50
44.27
34.25
Eau (%)
S
0.49
2.15
3.60
5.88
10.24
12.20
14.30
16.45
25.56
37.95
43.52
55.33
65.55
2. Tie line points
Aqueous Phase
Acetone
(mass %)
8.43
21.28
30.00
18.28
33.00
- 16 -
Organic Phase
Acetone
(mass %)
2.25
5.875
8.625
3.50
12.75